I'm Victoria — an Economics student and the founder of Axis Brief. Every week I break down one major shift at the intersection of AI and global power through the lens of economics. Not to inform you. To equip you.

“The cloud” is one of the most successful marketing terms in the history of technology. It suggests something weightless, distributed, everywhere and nowhere at once.

The reality is the opposite. Nearly every email, video call, financial transaction, and AI query that crosses an ocean travels through one of approximately 530 fiber optic cables lying on the seafloor — cables that are, in places, narrower than a garden hose. These cables carry an estimated 99% of intercontinental data traffic. Satellites — for all the attention they receive — carry a fraction of a percent.

The internet is not a cloud. It is hardware. Specific, physical, geographically located hardware that someone owns, someone built, and someone could damage.

Understanding who owns that hardware — and where it’s most vulnerable — is rapidly becoming one of the most consequential economic questions of the AI era.

The Geography That Cannot Be Changed

Undersea cables don’t connect continents directly. They funnel through chokepoints — narrow geographic corridors where the physical realities of ocean depth, coastlines, and existing infrastructure force dozens of cables into close proximity.

The Red Sea and the Bab-el-Mandeb Strait between Yemen and Djibouti carry an estimated 17 to 20 percent of global internet traffic between Europe, Africa, and Asia through a corridor a few dozen kilometers wide. In 2024, several major cables in this corridor were damaged — widely attributed to ship anchors in an area where Houthi attacks on shipping had already created instability. The economic disruption rippled through internet connectivity across the Middle East and East Africa for weeks.

The Singapore Strait performs a similar function for Asia-Pacific traffic — one of the densest concentrations of subsea cables on Earth, all passing through waters shared by some of the busiest commercial shipping lanes in the world.

The Taiwan Strait carries cables connecting East Asia’s major economies, in waters where commercial shipping, fishing fleets, and naval activity from multiple nations already overlap — and where cable damage incidents near Taiwan’s outlying islands have occurred multiple times in recent years, attributed variously to anchor drags and, in some cases, suspected deliberate activity.

The Baltic Sea has experienced a cluster of cable and pipeline damage incidents since 2023, prompting NATO members to increase naval patrols specifically to protect undersea infrastructure — an unprecedented peacetime allocation of military resources to protect fiber optic cables.

Here is the economic reality these incidents reveal: most cable damage is not sabotage. It is anchors, fishing equipment, and underwater landslides — mundane accidents that happen routinely. But because so many cables converge in so few corridors, even mundane accidents can have outsized consequences. And in regions where geopolitical tension is already elevated, the line between “accident” and “deniable gray-zone action” becomes economically and strategically blurry — which is itself a cost, regardless of what actually happened.

This geography cannot be re-engineered. The ocean floor’s depth contours, the location of continents, and decades of existing infrastructure mean new cables tend to follow similar routes to old ones. Chokepoints are not a temporary feature of the system. They are the system.

The Quiet Privatization Nobody Is Discussing

Here is the part of this story that receives almost no attention — and is, economically, the most significant.

For most of the history of undersea cables, ownership followed a consortium model. Multiple telecommunications companies — often dozens, sometimes representing different countries — would jointly fund a cable, share its capacity, and operate it as something closer to a shared utility than a private asset. This model reflected the economic reality that undersea cables were expensive, slow to build, and most valuable when shared broadly across many users.

That model is being replaced.

Google, Meta, Amazon, and Microsoft — the four largest cloud and AI infrastructure companies on Earth — have become the largest owners and lessors of undersea cable bandwidth in the world. Estimates suggest these four companies now own or control roughly half of all undersea cable bandwidth globally, a dramatic shift from a decade ago when telecom consortia dominated.

Some of this is leased capacity on consortium cables. Increasingly, it is wholly owned private cables — built by and for a single company, to move data between that company’s own data centers at the scale and latency its AI infrastructure demands.

Why does this matter economically? Because undersea cables were, for seventy years, infrastructure that approximated a public good — expensive to build but broadly shared, with governance structures involving multiple stakeholders and, often, multiple governments. As that infrastructure shifts toward private ownership by a handful of companies whose primary motivation is optimizing their own AI and cloud workloads, the economic incentives shift too.

A consortium cable’s economics depend on broad usage across many customers. A hyperscaler’s private cable economics depend on serving that hyperscaler’s own infrastructure most efficiently — which may or may not align with broader connectivity goals for the regions the cable passes through.

This is not a conspiracy. It is the rational economic response of companies whose AI infrastructure has outgrown what shared infrastructure can efficiently provide. But it represents a quiet transfer of control over the physical substrate of the global internet — from a broadly governed shared resource to the private infrastructure decisions of four companies headquartered in one country.

The China Dimension — Bifurcation Returns

If this pattern sounds familiar, it should. We’ve now seen it in semiconductors, in AI chip supply chains, and now in the physical cables that carry the data those chips process.

China’s HMN Technologies — spun out of Huawei’s marine networks division after Huawei faced Western sanctions — has rapidly become one of the world’s major cable-laying companies, central to China’s Digital Silk Road initiative connecting Chinese-built digital infrastructure across Africa, Southeast Asia, and Latin America.

In response, the United States and allied governments have increasingly pushed back on HMN-built cables landing in or near allied territory, citing espionage and data-security concerns — concerns that mirror, almost exactly, the language used around Huawei’s 5G equipment several years earlier. Alternative cable projects — backed by US, European, and allied capital and explicitly positioned as alternatives to Chinese-built infrastructure — have been announced across multiple regions.

The result is the same bifurcation pattern we examined in the semiconductor industry. Two parallel sets of physical infrastructure, built by different consortia, serving overlapping but increasingly separate digital ecosystems. A cable is, in this sense, not so different from a chip — it is infrastructure whose ownership and routing now carries strategic weight that has little to do with its technical function.

The Counterargument — Redundancy Is Real

The honest counterargument to all of this is that the industry knows about these vulnerabilities and is responding to them with real capital.

Investment in new subsea cable capacity is at record levels — tens of billions of dollars in announced projects, driven substantially by AI infrastructure demand. New routes are being designed specifically to avoid traditional chokepoints where possible. Mesh topologies — where data can reroute through multiple paths if one cable fails — are increasingly standard for new builds. Hyperscalers building private cables often build redundant pairs specifically to avoid single points of failure within their own networks.

This is genuine progress, and it matters. The internet of 2030 will likely be more resilient, in aggregate, than the internet of 2020.

But redundancy and chokepoints are not mutually exclusive. You can build ten new cables through the Red Sea corridor and the Red Sea corridor remains a chokepoint — you’ve simply raised the number of cables that would need to be damaged simultaneously to cause a major outage. Geography sets a floor under the risk that capital investment can reduce but cannot eliminate.

The most likely outcome by 2030 is the same pattern we keep encountering in this newsletter: meaningful mitigation, persistent underlying vulnerability, and a system that is more expensive and more resilient than before — but not fundamentally less concentrated at the points that matter most.

Three Economic Signals Worth Watching

Hyperscaler cable ownership disclosures. Google, Meta, Amazon, and Microsoft increasingly disclose their subsea cable investments in infrastructure announcements and sustainability reports. Track the share of new global cable capacity these four companies own outright versus lease — rising private ownership signals an accelerating shift away from the consortium model.

Chokepoint incident frequency and response time. Every cable damage incident in the Red Sea, Baltic Sea, and waters near Taiwan generates repair-time data. Repair times in contested waters have lengthened in recent years due to access and safety concerns. Lengthening repair times in any of these regions signal that geopolitical tension is beginning to directly degrade infrastructure resilience.

HMN Technologies contract wins versus Western alternative cable announcements. Every new cable project in Africa, Southeast Asia, or Latin America now functions as a small proxy contest between China’s Digital Silk Road and Western-backed alternatives. The win rate on each side over the next several years indicates how quickly the bifurcation pattern is spreading from semiconductors into physical internet infrastructure.

Follow the ownership filings. Not the ribbon-cutting ceremonies.

What This Means

The internet’s physical infrastructure has quietly become one of the most economically significant pieces of real estate on Earth — and almost nobody thinks about it, because “the cloud” was designed to make you not think about it.

Four companies now control roughly half the bandwidth that carries the world’s data between continents. A dozen narrow corridors on the ocean floor carry the overwhelming majority of that traffic. And the same bifurcation reshaping semiconductors is now reshaping the cables themselves.

None of this requires anything dramatic to matter. It requires only that the current trajectory continues — more AI demand, more private cable ownership, more geopolitical tension in the same handful of corridors that have carried this traffic for decades.

Most people will only think about undersea cables the day one of these corridors fails and a continent’s internet slows to a crawl.

Axis Brief exists so you understand the structure before that day arrives.

Next week: data centers — the new oil, and the most fought-over real estate of the AI economy.

— Victoria, Axis Brief

Leave a comment

I’m Victoria — an Economics student and the founder of Axis Brief. Every week I break down one major shift at the intersection of AI and global power through the lens of economics. Not to inform you. To equip you.

In 2014 China launched its first National Integrated Circuit Fund — known domestically as the Big Fund — with approximately $20 billion in state capital. The goal was straightforward: build a domestic semiconductor industry capable of reducing China’s dangerous dependence on foreign chips.

A decade later China has spent well over $150 billion across multiple Big Fund phases, with a third phase of approximately $47.5 billion launched in 2024. Dozens of fabs have been built. Thousands of engineers trained. Dozens of domestic chip designers funded.

And America’s response has been to conclude — correctly — that this represents an existential strategic threat, and to construct the most aggressive technology export control regime in modern history to slow it down.

Both countries are simultaneously right about what they’re doing. And both are missing something the economics makes clear.

What Money Can — and Cannot — Buy

The standard narrative around China’s semiconductor investment runs in one of two directions. Either the money is working and China is closing the gap faster than anyone admits. Or the money is wasted on corruption and inefficiency and the gap remains insurmountable.

Both narratives are wrong because they miss the fundamental economic reality of what semiconductors actually are.

Semiconductors are not a capital-intensive industry in the way steel or shipping is capital-intensive. They are a knowledge-intensive industry where capital is necessary but not sufficient. The difference matters enormously for understanding what $150 billion can accomplish.

Capital buys fabs. Capital buys equipment. Capital can poach talent and fund parallel research programs. What capital cannot buy — at any price, in any timeframe — is the accumulated tacit knowledge embedded in decades of manufacturing iteration.

Tacit knowledge is the economic concept for know-how that cannot be fully written down or transferred through instruction. It lives in the hands and instincts of engineers who have spent years optimizing specific processes. It lives in the supplier relationships built through thousands of iterations of problem-solving. It lives in the institutional memory of organizations that have debugged the same class of problem seventeen times across fifteen years.

TSMC’s manufacturing excellence is not primarily a function of its equipment or its funding. It is a function of forty years of compounding tacit knowledge. When TSMC builds a new facility in Arizona, it ships its own engineers to transfer that knowledge. Even then — as the Arizona facility’s higher costs relative to Taiwan demonstrate — the full ecosystem cannot be perfectly replicated even by TSMC itself.

China’s $150 billion is buying real and significant progress. It is not buying what it most needs: time. And time — measured in decades of manufacturing iteration — cannot be purchased.

What China Has Actually Built

The honest accounting of China’s semiconductor progress is more nuanced than most Western coverage acknowledges — and more limited than Chinese state media claims.

On the genuine progress side: SMIC — China’s most advanced chip manufacturer — has achieved 7nm-class production through a technique called multi-patterning DUV. This means using older lithography equipment in multiple passes to achieve resolutions that normally require more advanced machines. It is slower, more expensive, and lower-yield than leading-edge production. But it works. Huawei’s Mate 60 Pro — launched in 2023 to considerable surprise — runs on a SMIC-produced chip that demonstrates this capability is real.

China’s overall fab capacity is expanding rapidly. Independent analysts project China’s share of global semiconductor manufacturing capacity reaching approximately 20 to 21 percent by 2030. In legacy and mid-tier chips — the semiconductors used in automobiles, consumer electronics, industrial equipment, and most non-AI applications — China is advancing toward meaningful self-sufficiency. Targets of 50 percent or higher overall self-sufficiency in volume terms by the late 2020s appear achievable.

On the persistent limits: The fundamental chokepoint remains EUV lithography. ASML — the Dutch company that produces the extreme ultraviolet lithography machines required for cutting-edge chip production below 5nm — is effectively a monopoly supplier. Export controls coordinated between the United States, Netherlands, and Japan have blocked ASML from selling its most advanced equipment to China since 2019. China cannot currently produce EUV machines domestically at commercial scale. Prototype efforts exist. Commercial viability remains years — possibly decades — away.

Beyond EUV: photoresists, precision optics, electronic design automation software, and the specialized equipment ecosystem that surrounds leading-edge production are predominantly American, Dutch, and Japanese. Each represents a chokepoint. Collectively they define a frontier that China’s capital alone cannot cross quickly.

The Solar Panel Parallel Nobody Is Drawing

Here is the economic insight that most semiconductor analysis misses entirely.

China does not need to match TSMC at the frontier to fundamentally reshape global power dynamics. It needs to achieve something far more achievable: dominance in the mid-tier.

We have seen this movie before. Twice.

In solar panels China entered a market it did not lead, deployed massive state subsidies to scale production, drove costs below what any competitor could match, and effectively ended the solar manufacturing industries of Germany, the United States, and most of Europe. Today China produces approximately 80 percent of the world’s solar panels. The frontier of solar technology still has innovators in multiple countries. The economics of production belong entirely to China.

In electric vehicles the same dynamic is playing out in real time. Chinese EV manufacturers — backed by state subsidies, domestic market scale, and vertically integrated supply chains — are producing vehicles at price points that Western manufacturers structurally cannot match without equivalent subsidies. The technological frontier remains contested. The cost economics are moving decisively toward China.

The semiconductor analogy is not perfect — chips are more complex, the knowledge barriers are higher, the equipment dependencies are more acute. But the strategic logic is identical. China does not need to produce the world’s most advanced chips. It needs to produce enough chips, at competitive enough cost, to supply its own economy and flood global markets for everything below the frontier.

If that happens — and the trajectory suggests it is happening in legacy and mid-tier segments — America’s leverage evaporates in precisely the domains where most of the world’s actual semiconductor demand lives.

The Trap Partially Backfires

America’s export control strategy is working. And it is failing. At the same time.

It is working in the narrow sense that it has genuinely slowed China’s progress toward the frontier. Without access to EUV lithography and the broader equipment ecosystem, China cannot close the leading-edge gap on the timeline it originally intended. The controls have bought time — measured in years — for American and allied investment in domestic production and next-generation technology.

It is failing in a broader sense that the controls have accelerated exactly the strategic outcome they were designed to prevent. By making clear that America views semiconductor technology as a weapon of strategic competition, Washington has given China’s leadership an undeniable justification for the massive state investment required to reduce dependence. Before the controls, China’s semiconductor ambitions faced internal resistance from economists who questioned the efficiency of state-directed industrial policy at this scale. After the controls, that debate was settled.

The more durable failure is structural. Export controls leak over time. Smuggling routes emerge. Third-country intermediaries develop. Domestic alternatives improve. Chinese engineers reverse-engineer what they cannot buy. The controls impose costs and slow timelines — they do not stop the trajectory.

What emerges is not Chinese semiconductor dominance or American semiconductor security. It is bifurcation — two parallel semiconductor ecosystems, one centered on American and allied technology, one increasingly centered on Chinese domestic production. A Silicon Curtain, to extend the Cold War metaphor.

Bifurcation is more expensive than integration for everyone. It means duplicated research, duplicated production capacity, higher costs, and slower innovation from reduced collaboration. The world pays a premium for the geopolitical anxiety of two great powers that no longer trust shared infrastructure.

The Deepest Economic Question

Strip away the geopolitics and China’s $150 billion bet poses a question that economists have debated for decades but never seen tested at this scale:

Can concentrated state capital overcome network effects and tacit knowledge in the most complex manufacturing system human civilization has ever built?

The historical evidence is mixed. State-directed industrial policy has succeeded in catching up in commoditized, capital-intensive industries — steel, shipbuilding, solar, EVs. It has struggled at the knowledge-intensive frontier where creativity, risk-taking, and cross-border collaboration define leadership.

Semiconductors sit at the extreme knowledge-intensive end of that spectrum. The evidence so far suggests China’s investment is succeeding in exactly the domains where state-directed policy historically succeeds — scale, capacity, cost reduction in established processes — and struggling in exactly the domains where it historically struggles — frontier innovation, yield optimization, ecosystem development.

That pattern suggests a durable partial outcome: China achieves meaningful self-sufficiency in non-cutting-edge semiconductors, reduces American leverage over its economy, and creates a resilient if less efficient domestic ecosystem. It does not match TSMC or NVIDIA at the frontier within any near-term horizon.

For America, that outcome is better than Chinese semiconductor supremacy and worse than sustained Chinese dependence. It is also the most likely outcome. Policy built around preventing it entirely — rather than managing it — is building on a foundation that the economics do not support.

Three Economic Signals Worth Watching

SMIC yield rates at advanced nodes. Production capability and production at scale are different things. SMIC can produce 7nm-class chips. Whether it can do so at yields — the percentage of functional chips per wafer — that make mass production economically viable is the question that determines strategic significance. Yield data is rarely published. Infer it from customer adoption rates and pricing.

Mature node overcapacity pricing. If China’s mid-tier semiconductor expansion creates the same dynamic as solar and EVs — prices falling below the cost of production for non-subsidized competitors — watch for Western and Taiwanese manufacturers beginning to exit or consolidate in legacy chip segments. That exit is the signal that the trap has partially backfired.

Big Fund III allocation transparency. The third phase of China’s National IC Fund has tighter oversight mechanisms following corruption investigations that compromised earlier phases. How that capital is actually allocated — between frontier moonshots and pragmatic capacity expansion — reveals whether China has learned the industrial policy lessons of its first decade of semiconductor investment.

Follow the yield data. Not the fab announcements.

What This Means

China’s $150 billion bet will not produce semiconductor independence. It will produce something strategically more significant: semiconductor resilience sufficient to make America’s leverage increasingly expensive to maintain.

The world is not moving toward Chinese chip supremacy or American chip security. It is moving toward two parallel systems — more expensive, less efficient, more tense — where the cost of geopolitical anxiety gets distributed across every technology product every consumer on earth buys.

That cost is real. It is already being paid. And it will compound for as long as the competition continues.

Most people think of the semiconductor race as a story about who builds the best chips.

The economics suggest it was always a story about who can afford to stop depending on someone else.

Axis Brief exists to show you the difference.

Next week: Who controls the undersea cables controls the internet — and the race to own that infrastructure is already decided.

— Victoria, Axis Brief

Leave a comment

I'm Victoria — an Economics student and the founder of Axis Brief. Every week I break down one major shift at the intersection of AI and global power through the lens of economics. Not to inform you. To equip you.

There is a concept in finance called an uninsurable risk.

It’s not a risk that’s unlikely. It’s a risk so large, so interconnected, and so catastrophic that no insurance market can price it. No premium is high enough. No payout would cover the damage. The risk simply sits there — acknowledged, unhedged, growing — while the world builds more and more around it.

Taiwan is the largest uninsurable risk in the modern global economy.

Not because of its military position. Not because of its political status. But because of what sits inside its factories — and what disappears from the global economy the moment those factories stop working.

What TSMC Actually Is — Economically

Most coverage of Taiwan frames TSMC as a very important company. That framing understates the reality by an order of magnitude.

TSMC is not a company in the conventional sense. It is a natural monopoly built over four decades of deliberate, compounding investment — in technology, in talent, in supplier relationships, in accumulated manufacturing knowledge that cannot be written down or replicated quickly regardless of how much capital you throw at the problem.

Natural monopolies don’t emerge from luck. They emerge from path dependency — the economic phenomenon where early advantages compound into structural dominance that becomes progressively harder to challenge over time. TSMC’s early lead in advanced chip manufacturing created an ecosystem around it. Specialized equipment makers, chemical suppliers, engineering talent, process knowledge. Each addition to that ecosystem made TSMC more productive. Each productivity gain attracted more customers. Each customer relationship funded more research. The cycle repeated for decades.

The result: TSMC produces over 90% of the world’s most advanced logic chips — the semiconductors below 7 nanometers that power AI systems, high performance computing, modern smartphones, automotive intelligence, and defense systems. Every American AI company. Every Chinese AI ambition. Every autonomous military system in development anywhere on earth. All of them depend on chips predominantly made in one location.

I’ll be honest — the number I keep returning to is not 90%. It’s the implication of 90%. If a single supplier controls 90% of any critical input in any industry — food, energy, water, medicine — economists immediately flag it as a systemic vulnerability requiring urgent diversification. When that input is the foundational component of the entire AI and technology economy, the urgency should be proportionally larger.

It is not.

The Silicon Shield — And Why It’s a Trap

Taiwan’s strategic position is often described as protected by the “silicon shield” — the idea that its semiconductor dominance deters military conflict because attacking Taiwan would cause catastrophic economic damage to the attacker.

The logic is sound. China relies heavily on Taiwanese chips. A military takeover would not instantly transfer TSMC’s capabilities to Beijing — the expertise, the supply chains, the institutional knowledge embedded in thousands of engineers and processes cannot be seized like territory. Destroying Taiwan’s fabs would cripple China’s own technology sector as surely as anyone else’s. The silicon shield creates a version of mutual assured economic destruction.

But here is the economic problem with mutual assured destruction as a deterrent: it only works when both sides are rational, fully informed, and operating under stable conditions.

It does not account for miscalculation. It does not account for a leadership that concludes the economic damage is survivable or temporary. It does not account for a crisis that escalates beyond anyone’s original intentions. And critically — unlike nuclear deterrence which produced decades of arms control frameworks — there is no international architecture managing the Taiwan semiconductor risk. No treaty. No agreed red line. No mechanism for de-escalation built specifically around economic interdependence.

The silicon shield is not a security guarantee. It is a bet that rational economic incentives will always override political and military impulses.

History does not strongly support that bet.

The CHIPS Act: Necessary Hedge, Not Full Solution

In 2022 the United States passed the CHIPS and Science Act — $52 billion in subsidies to rebuild domestic semiconductor manufacturing. The political logic was clear: reduce dependence on Taiwan by bringing advanced chip production to American soil.

The honest assessment of where that effort stands is more nuanced than either its advocates or critics typically acknowledge.

The CHIPS Act is working better than many expected on measurable outputs. Hundreds of billions in private investment have been announced. Construction has begun on major facilities in Arizona, Ohio, and Texas. Jobs are being created. The ecosystem of suppliers and talent beginning to form around these facilities is real.

But strategically, it is a necessary hedge — not a full solution — and almost certainly not within a five year horizon.

Here is why. TSMC’s competitive advantage is not primarily its equipment or its funding. It is the decades of accumulated process knowledge embedded in its workforce and operations. TSMC’s Arizona facility — built with TSMC’s own engineers and technology — is producing chips, but at significantly higher cost than its Taiwan operations. The gap reflects not inefficiency but the absence of the full ecosystem that makes Taiwanese production uniquely productive.

Rebuilding that ecosystem in America takes time measured in decades, not years. The CHIPS Act accelerates the process. It does not compress it into a strategic timeframe that addresses near-term Taiwan risk.

What the CHIPS Act does accomplish — and this matters — is begin shifting the trajectory. Every year of sustained investment narrows the gap. Every facility built reduces the concentration of risk. The CHIPS Act is not a solution. It is the beginning of a solution that requires sustained political will across multiple administrations to complete.

That last condition is the most economically uncertain of all.

The Uninsurable Tail Risk

Here is what makes Taiwan genuinely different from every other geopolitical risk in the global economy.

You can insure against supply chain disruption. You can insure against political risk in emerging markets. You can hedge against currency crises, commodity shocks, regulatory changes. The insurance and derivatives markets exist precisely to price and distribute these risks across participants who can bear them.

Nobody can insure against losing Taiwan.

Not because the risk is unquantifiable — economists have attempted to estimate the economic cost of a Taiwan conflict. The range runs from $2 trillion to $10 trillion in direct economic damage in the first year, with cascading effects that compound over subsequent years as the technology sector adjusts to zero advanced chip production. Those numbers are staggering enough.

The deeper problem is structural. Insurance works by pooling risk across many independent events. Taiwan is a single point of failure for the entire global technology economy. There is no pool large enough, no counterparty solvent enough, to make that risk insurable.

This means every company, every government, every investor with technology exposure carries Taiwan risk on their balance sheet — unhedged, unpriced, and largely unacknowledged. The global technology sector has built $30 trillion in market value on infrastructure that has no insurance policy.

That is an extraordinary economic fact that receives far less attention than it deserves.

Why China Can’t Simply Take What It Wants

This is the economic reality that makes the Taiwan situation genuinely complex rather than a simple story of Chinese aggression.

China’s own technology sector runs on Taiwanese chips. Huawei, despite years of American sanctions pressure, relies on chips processed through TSMC’s supply chain. Chinese AI development depends on the same foundational semiconductor infrastructure as everyone else. A military action that destroyed or even significantly disrupted TSMC’s operations would inflict severe damage on China’s own technology ambitions — ambitions that Xi Jinping has placed at the center of China’s development strategy.

This creates a paradox. China wants Taiwan — for historical, political, and strategic reasons. But China also needs Taiwan’s chips — for economic and technological reasons. Taking Taiwan militarily risks destroying the very asset that makes it valuable.

The most sophisticated Chinese economic planners understand this. Their semiconductor self-sufficiency drive — the $150 billion investment we examined last week — is partly about reducing this vulnerability before any potential military action becomes economically viable. China is trying to build itself a path where it could act without destroying what it needs.

Whether that path is achievable in the relevant timeframe is the most important strategic question in global economics right now. Most independent analysts believe it is not — that China cannot close the semiconductor gap fast enough to make military action economically rational within the current decade.

But “not economically rational” and “won’t happen” are different statements.

Three Economic Signals Worth Watching

TSMC Arizona cost convergence. The gap between production costs at TSMC’s Arizona facility and its Taiwan operations is the single best indicator of how quickly genuine diversification is occurring. As that gap narrows — if it narrows — the economic case for Taiwan’s irreplaceability weakens. Watch TSMC’s quarterly earnings calls for any mention of yield rates and cost parity.

Chinese SMIC advanced node announcements. Every time SMIC announces production capability at a smaller node size, China’s semiconductor self-sufficiency timeline shifts. The gap between announcement and reliable mass production is typically large — but the direction of travel matters. Watch for 5nm claims and scrutinize them carefully.

CHIPS Act facility completion timelines. The distance between announced timelines and actual production dates at American facilities is the most honest measure of how quickly domestic diversification is actually proceeding. Intel’s Ohio facility has already slipped its original timeline. Track the slippage.

Follow the production yields. Not the press conferences.

What This Means

Taiwan’s centrality in the global technology economy is not a geopolitical problem with a diplomatic solution. It is an economic problem with no clean solution — only risk management, diversification, and the slow, expensive work of rebuilding supply chain resilience over years and decades.

The world built a $10 trillion dependency on the productive capacity of 23 million people living on an island at the center of the most contested body of water on earth. That dependency did not emerge from negligence. It emerged from decades of rational economic decisions — each individually sensible, collectively creating a systemic vulnerability that no single actor can fix.

What makes it genuinely dangerous is not the possibility of conflict. It is the absence of any mechanism — financial, diplomatic, or institutional — capable of managing the risk if conflict occurs.

The world has built its technological future on an uninsured foundation.

Most people will only understand what that means after the foundation cracks.

Axis Brief exists so you understand it before.

Next week: Who controls the undersea cables controls the internet — and the race to own that infrastructure is already decided.

— Victoria, Axis Brief

I’m Victoria — an Economics student and the founder of Axis Brief. Every week I break down one major development at the intersection of AI and global power through the lens of economics. Not to inform you. To equip you.

Everyone is watching America and China.

That’s exactly what Singapore, Vietnam, and the UAE want.

While the world’s attention is fixed on two superpowers competing for dominance, three smaller economies have quietly built something more valuable than allegiance to either side. They’ve built the infrastructure both sides need — and they’re charging accordingly.

This isn’t neutrality. Neutrality is passive. What Singapore, Vietnam, and the UAE are doing is something more deliberate and more profitable.

They made themselves impossible to replace.

The Economics of Being Indispensable

There is a concept in economics called bilateral monopoly — a situation where a single seller faces a single buyer and both sides need each other enough that neither can walk away. The seller extracts maximum value not by competing but by being the only viable option.

Singapore, Vietnam, and the UAE have each applied a version of this logic to geopolitics.

They aren’t selling allegiance. They’re selling access — to capital markets, to manufacturing capacity, to digital infrastructure. And because both America and China need that access, both sides pay for it.

The question worth asking is not whether this strategy is clever. It clearly is. The question is how long it holds — and what it costs to maintain it as the pressure to choose sides increases.

Singapore: The Toll Booth at the World’s Most Important Intersection

Singapore’s GDP is roughly $500 billion. Its geographic footprint is smaller than many cities. Yet it hosts the regional headquarters of more Fortune 500 companies than almost any other city on earth. It processes a disproportionate share of Asian financial flows. And in the last three years it has become one of the fastest-growing AI and data center hubs in the world.

None of this is accidental.

Singapore understood something early that most countries are still figuring out: the most valuable economic position in a bipolar world is not alignment with the stronger side. It’s ownership of the infrastructure both sides need to function.

American banks need Singapore to access Asian capital markets. Chinese firms need Singapore to access Western credibility — to structure deals, raise international funding, and operate within legal frameworks that global investors trust. Singapore provides both services simultaneously and charges both sides for the privilege.

I’ll be honest — when I first looked at Singapore’s foreign direct investment numbers I assumed I was misreading them. Singapore — a city of six million people — consistently attracts more FDI than countries fifty times its size. The explanation isn’t geography. It’s positioning.

What Singapore has built is essentially a toll booth at the most important economic intersection in the world. Every major transaction between East and West passes through it. Every significant capital flow touches it. And unlike a physical toll booth, this one becomes more valuable as traffic increases — not less.

The AI race is increasing traffic.

As U.S. and Chinese tech firms race to build data centers, secure supply chains, and establish regional operations, Singapore keeps appearing as the preferred location. It offers political stability, rule of law, deep talent pools, and — critically — a government that has never forced a foreign company to choose between its American and Chinese relationships.

That last point is worth more than any tax incentive.

Vietnam: The $100 Billion Pivot Nobody Planned

In 2010 Vietnam exported approximately $12 billion worth of goods to the United States. By 2023 that number exceeded $100 billion.

That is not organic growth. That is a structural shift in global manufacturing — and Vietnam was positioned to capture it almost perfectly.

Here is what happened economically. As U.S.-China trade tensions escalated through the 2010s and accelerated after 2018, American companies began executing what became known as the China+1 strategy — keeping Chinese operations while establishing a second manufacturing base outside China to reduce tariff exposure and supply chain concentration risk.

Vietnam was the primary beneficiary.

Samsung moved significant portions of its smartphone production to Vietnam. Intel built one of its largest assembly and test facilities there. LG, Apple’s supplier network, and dozens of other major manufacturers followed. Vietnamese electronics exports — the category you correctly identified as the core — grew from a negligible base to over $100 billion annually, making it one of the fastest manufacturing scale-ups in modern economic history.

But here’s the economic nuance that most coverage misses.

Vietnam didn’t just attract manufacturing that left China. It attracted manufacturing that still depends heavily on Chinese inputs. Semiconductors, components, raw materials — much of what gets assembled in Vietnam begins its journey in China. Vietnamese factories finish the product. Vietnamese export figures record the full value. American importers pay Vietnamese tariff rates.

This arrangement benefits Vietnam enormously. It also means Vietnam’s economic relationship with China remains deep even as its security relationship tilts toward the United States.

Vietnam upgraded its ties with America to a Comprehensive Strategic Partnership in 2023 — the highest diplomatic designation Washington offers. That same year Vietnam’s trade with China hit record levels.

That is not contradiction. That is precision. Vietnam has correctly identified that its economic interests require China’s supply chains and America’s consumer markets simultaneously. Disrupting either relationship costs more than maintaining both.

The balancing act is expensive to sustain. Vietnam walks a narrower path than Singapore — it has less financial cushion, less institutional depth, and a more complicated historical relationship with its largest neighbor. But the economic returns so far have been significant enough to make the tightrope worth walking.

UAE: Buying Sovereignty Before the Window Closes

The UAE’s AI strategy is the most ambitious of the three — and the most economically revealing.

Since 2020 UAE sovereign wealth funds and government entities have committed tens of billions of dollars to AI infrastructure. G42 — Abu Dhabi’s state-linked AI company — has built partnerships with Microsoft, OpenAI, and several Chinese technology firms simultaneously. The UAE has positioned itself as a major data center hub, attracting hyperscaler investment from both American and Chinese cloud providers. And in 2024 the UAE secured a landmark agreement with the United States for access to advanced AI chips — a deal that reflects how seriously Washington takes Abu Dhabi’s emerging role in global AI infrastructure.

To understand why the UAE is doing this you have to understand what the UAE learned from oil.

For decades the Gulf states wielded extraordinary global influence because the world needed their energy. That influence was real but it came with a structural vulnerability — it depended entirely on a single resource that the world was slowly but deliberately moving away from. The UAE watched as the energy transition accelerated and drew the correct economic conclusion: resource dependence is leverage until it isn’t.

The AI infrastructure play is the UAE’s answer to that lesson.

If the UAE successfully builds sovereign AI capability — its own models, its own data centers, its own digital infrastructure that doesn’t depend on American or Chinese systems — it creates something no Gulf state has ever had before. The ability to set its own terms in every major economic negotiation for the next fifty years. Not because it has oil. But because it has the infrastructure others need.

Your instinct about what UAE looks like in 2040 was precise: true independence with its own market. The economic translation of that is a country that cannot be sanctioned into compliance, cannot be technologically isolated, and cannot be forced to choose between Washington and Beijing because it built a third option.

The Third Option Changes Everything

This is where the economics become genuinely consequential.

Right now most of the world faces a binary choice in AI infrastructure. American systems — built on U.S. chips, governed by U.S. law, accessible to U.S. intelligence — or Chinese systems, with their own dependencies and their own risks. Countries that choose American AI accept certain constraints. Countries that choose Chinese AI accept different ones. There is no neutral ground.

Until there is.

If the UAE — backed by Gulf capital, positioned between East and West, with partnerships on both sides — successfully builds credible sovereign AI infrastructure, the binary breaks. Suddenly the world has a third option. And the moment a credible third option exists, the leverage that both America and China currently hold over every other nation weakens permanently.

Neither Washington nor Beijing is publicly accounting for this possibility. Both are focused on each other.

That is a significant strategic oversight.

Countries that have spent decades being pawns in great power competition are watching the UAE’s experiment closely. If it works — if genuine AI sovereignty can be purchased with enough capital and enough diplomatic skill — the lesson spreads. Other nations start asking whether they can build their own version. The bilateral monopoly that America and China currently enjoy over global AI infrastructure starts to erode.

That’s not just an economic shift. That’s a restructuring of global power that changes the terms of every major negotiation for the next half century.

Three Economic Signals Worth Watching

G42’s partnership decisions. Abu Dhabi’s state AI company currently maintains relationships with both American and Chinese technology firms — an unusual position that Washington has already pushed back on. Watch which relationships G42 expands and which it scales back. Those decisions will tell you whether UAE sovereignty is real or performative.

Vietnamese electronics export composition. The ratio of Chinese-input manufacturing to genuinely domestic Vietnamese production is the number that reveals how sustainable Vietnam’s balancing act is. As that ratio shifts — either direction — the strategic position shifts with it.

Singapore’s data center capacity growth. Singapore has actually imposed temporary limits on new data center construction due to energy constraints. How it manages that constraint — who gets capacity and who doesn’t — reveals exactly how it prices access in a world where both superpowers want the same real estate.

Follow the investment flows. Not the diplomatic statements.

What This Means

The AI race is most commonly described as a contest between two powers. That framing is incomplete.

It is a contest being shaped by dozens of smaller economies making calculated decisions about where to position themselves — decisions that will determine the architecture of global trade, technology, and power for the next generation.

Singapore, Vietnam, and the UAE are not bystanders. They are active participants extracting maximum economic value from a competition they did not start and cannot control. The strategies are different. The underlying logic is identical.

Make yourself impossible to replace. Charge accordingly. Keep both sides competing for your partnership.

Most of the world is still figuring out that this is an option.

The ones who figured it out first are already getting paid.

Next week: Why Taiwan is the most economically consequential piece of real estate on earth right now — and it has nothing to do with politics.

— Victoria, Axis Brief

I'm Victoria — an Economics student and the founder of Axis Brief. I started this newsletter because the most consequential competition in modern history was being covered like a sports match — scores and highlights, no analysis of what's actually at stake. Every week I break down one major development in AI and global power through the lens of economics. What follows is what I found this week.

Let me say something the major outlets are dancing around.

America is not blocking chip exports to China because it’s winning. It’s doing it because it’s falling behind in one critical dimension of this race — and buying time is the only lever it has left to pull.

Washington just bought itself roughly five years. Nobody in power is publicly explaining what happens when those five years run out. That silence is more telling than anything being said.

A Pattern Nobody Wants To Acknowledge

In 1941 the United States cut off Japan’s oil supply. The stated reason was economic pressure. The real reason was strategic — Japan was expanding across Asia faster than America could respond diplomatically and something had to slow it down.

Within six months Japan attacked Pearl Harbor.

I’m not drawing this parallel to be dramatic. I’m drawing it because the strategic logic is strikingly similar to what’s happening right now — and ignoring historical patterns because they’re uncomfortable is exactly how people get blindsided by what comes next.

Technological embargo is rarely just an economic tool. At a certain scale it becomes a signal of strategic urgency. The resource changes — oil then, semiconductors now — but the underlying calculation is familiar: constrain a rising power’s access to what it needs most before the gap becomes impossible to close.

The question worth sitting with is: how did that work out in 1941?

What October 2022 Actually Was

The U.S. government told its most profitable companies they could no longer sell advanced chips to China. No exceptions. No negotiation. Not without explicit government permission.

The headlines called it a trade restriction. That framing is incomplete and it matters that it’s incomplete.

Trade restrictions are designed to change economic behavior. What America announced in October 2022 was designed to slow a specific technological outcome regardless of economic cost. American companies lost billions in Chinese contracts overnight and Washington accepted that loss without flinching.

That is not trade policy. That is strategic denial — a concept more at home in military doctrine than economic policy.

Now here’s the part that deserves more attention than it’s getting.

TSMC — a single Taiwanese company — manufactures the overwhelming majority of the world’s most advanced semiconductors. Estimates consistently place its share of cutting-edge chip production above 90%. Every major American AI company. Every Chinese AI ambition. Every advanced weapons program being developed anywhere on earth. All of it depends on chips predominantly made in one place — an island of 23 million people sitting 100 miles off the coast of China.

I’ll be honest — when I first fully understood what that concentration actually meant I sat with it for a while. One company. One island. The entire race. If something happens to Taiwan — politically, militarily, or even from a natural disaster — the global AI competition doesn’t pause. It collapses and restarts from near zero.

That is an extraordinary single point of concentration in the history of modern technological civilization. It receives a fraction of the attention it deserves.

And here’s the detail that made it land harder for me. TSMC’s founder Morris Chang said publicly in 2022 that globalization is dead. Not weakening. Not under pressure. Dead. The man who built the company the entire world’s AI future runs through looked at what was happening and concluded the era of interdependent global supply chains — the economic foundation of seventy years of relative peace — is finished.

When the architect of the system tells you the system is broken, the correct response is not to wait for a second opinion.

Why The Ban May Not Be Enough

This is where the analysis gets uncomfortable for any simple narrative — American or Chinese.

China has committed over $150 billion to building its own domestic semiconductor industry. SMIC — China’s state chip manufacturer — has already demonstrated the ability to produce advanced chips despite the export ban, though still trailing the leading edge. Independent analysts estimate China could meaningfully close the technology gap within the next several years — some placing that window as early as 2027, others more conservatively.

The ban slowed China. Whether it stopped China is a different question — and the evidence suggests the answer is no.

So let’s be precise about what October 2022 actually accomplished: it bought time. The clock started running and Washington has not publicly explained what it intends to build during that window. Maybe there is a comprehensive plan. Maybe it’s classified. Maybe the CHIPS Act investments and export controls together constitute the strategy and we’re watching it unfold.

What’s clear is this — tariffs and export bans alone don’t restore industrial capacity. They create breathing room. Breathing room only matters if you use it.

Where China Has A Genuine Lead

Walk into a convenience store in Shenzhen and your face pays for your coffee. Board a train in Shanghai and AI has already optimized every departure across the entire national rail network. Check into a hospital in Beijing and an algorithm has begun your preliminary diagnosis before a doctor enters the room.

This is not science fiction. This is daily life in China’s major cities right now.

China processes more mobile digital payments than the United States and Europe combined. Its AI patent filings have surpassed America’s in volume. The gap in AI deployment into everyday infrastructure — payments, transport, healthcare, surveillance — is real and significant.

Here is what that means in economic terms. When AI becomes embedded in the daily behavior of 1.4 billion people — their payments, their transport, their healthcare, their education — it creates switching costs and data advantages that no export ban can touch. The infrastructure is already built. The habits are already formed. The data is already accumulated.

This is China’s genuine advantage and it’s important to name it precisely — not as overall AI supremacy, but as deployment scale and integration depth that the U.S. has not matched domestically.

The U.S. retains significant countervailing advantages — leading AI research institutions, the dollar as global reserve currency, energy independence, and the world’s most powerful alliance network. These are real and they matter. The chip restrictions represent America leveraging its strongest remaining card — supply chain control — precisely because the deployment race has gone in a different direction.

The Military Problem Nobody Is Pricing In

Both countries are pouring capital into autonomous military systems. Drones that make targeting decisions. Algorithms that identify threats faster than human perception. Battlefield tools that operate in the milliseconds between a human thought and a human action.

The economics of this are coldly straightforward. Whoever automates military decision-making effectively reduces the cost of projecting power. Lower cost of projection means more power projected across more situations.

We have been here before in terms of the strategic logic. In 1945 two bombs detonated and the world immediately grasped what was at stake. Within years every major power was racing to build nuclear weapons. Within two decades the world had constructed the Nuclear Non-Proliferation Treaty — an imperfect but functional legal architecture for managing existential risk.

We are at an early equivalent moment for autonomous weapons. The most dramatic demonstrations have not yet occurred. The race is already running. And unlike nuclear weapons — which require rare materials and massive visible infrastructure — autonomous systems can be developed more quietly and at greater scale.

There is currently no binding international framework governing autonomous military AI. No treaty. No agreed red line. Not because nobody thought of it — but because both the U.S. and China are deliberately avoiding commitment. In game theory terms both players face a classic prisoner’s dilemma — neither wants to constrain themselves first, both face worse outcomes if neither does. The rest of the world absorbs the cost of that standoff.

The Strategic Logic On Both Sides

Strip the politics away and two very old strategies are playing out with new tools.

China is competing on deployment scale — embed AI into infrastructure faster than anyone else and create dependencies so deep that the rest of the world integrates with Chinese systems by practical necessity rather than deliberate choice. This echoes how Britain built its empire — not primarily through conquest but through trade relationships that became structural dependencies. By the time countries fully understood how locked in they were the cost of exit exceeded the cost of staying.

America is competing on foundational control — own the critical technology layer and maintain leverage over everyone who needs it to build anything. This resembles the Standard Oil model applied to semiconductors. Control the essential input and you retain influence over the output regardless of who does the manufacturing.

Both strategies have historical precedent. Both have costs that only become fully visible decades later. Both are being pursued simultaneously by nuclear-armed states with fundamentally different visions of what the global order should look like.

That combination warrants serious attention.

Three Signals Worth Watching

Most geopolitical coverage tells you what already happened. Here are three forward-looking economic indicators worth tracking:

TSMC capital expenditure decisions. When Taiwan’s dominant chip manufacturer shifts where it builds new fabrication facilities a major power just made a major move. The Arizona announcement in 2022 was not driven by market economics alone — the subsidies required made that clear. It was a geopolitical decision structured as a corporate one.

AI infrastructure investment across Southeast Asia. Vietnam, Indonesia, Malaysia, and the Philippines are receiving significant investment from both Washington and Beijing in data centers, undersea cables, and chip assembly. The alignment decisions these countries make over the next several years will shape the architecture of global technology infrastructure through 2040 and beyond. This story is significantly underreported.

Defense budget allocations for autonomous systems. The U.S. has dramatically increased spending on autonomous weapons development in recent years — public figures run into the billions annually across services. China does not publish equivalent granular figures. When detailed defense spending data becomes less transparent the standard analytical assumption is that the program expanded rather than contracted.

Follow the capital allocation. Not the press releases.

Why This Matters

The competition between the United States and China over AI and semiconductors will influence which currencies and legal frameworks dominate global commerce, which military doctrines shape the next generation of conflict, and which values get embedded into infrastructure that billions of people use daily.

The honest picture is neither American fatalism nor Chinese triumphalism. It is two powerful countries with different structural advantages pursuing incompatible visions of global order through economic and technological competition — with the rules still being written and the outcome genuinely uncertain.

Most people are watching this the way you watch weather — as something that happens to them, decided elsewhere, by people they’ll never meet.

The decisions being made right now will shape the world your generation inherits. A small number of people are paying close enough attention to understand what’s actually being decided while it’s still being decided.

That gap — between those who see it and those who don’t — is the most consequential intellectual divide of our generation.

Axis Brief exists to close it.

Next week: The small nations positioning themselves between these two giants — and why their economic choices will determine the outcome of a race they’re not officially running.

— Victoria, Axis Brief